“…For , r ≥ 3, Győri and Lemons [13] proved ex(n, C r ) = O(n 1+ 1 /2 ), so up to the o(1) term, ( * ) would be best possible. This conjecture is known to hold for = 3 and all r due to work of Ruzsa and Szemeredi [24] and Erdős, Frankl, and Rödl [5], and for = 4 and all r due to Lazebnik and the second author [16] and Timmons and the second author [27]. We emphasize that for r ≥ 3, it is known that the o(1) term in ( * ) is necessary in general, see Ruzsa and Szemeredi [24] and Conlon, Fox, Sudakov, and Zhao [3].…”